Introduction
OpenAI-compatible AI gateway with Turkish personal data masking. Send your first request in minutes.
Gurubase Siper consists of three parts: an OpenAI-compatible AI gateway, personal data masking for Turkish text, and a management panel. Requests are masked in Turkey before they go to the external language model. It is designed for teams that work with text containing personal data and do not want to send that data as-is to models abroad.
In practice, masking looks like this:
Start here
Getting started
Get your virtual key and send your first request with the OpenAI SDK.
Concepts
How masking works, the placeholder format, fail-closed behavior, and data sovereignty.
Categories
The 26 masked personal data categories, with synthetic examples.
Guides
Switching with the OpenAI SDK, streaming, and mask-only usage.
API reference
Full reference for the Gateway and masking endpoints.
Security
Data retention approach, audit trail, and access controls.
To create an account, contact us; if your account is ready, you can sign in to the panel and generate your virtual key.
Siper is part of the Gurubase product family. Gurubase lets you build AI agents that draw on your organization’s own knowledge sources, cite a source in every answer, and decline to answer when the context is weak; Siper is the family’s Turkish personal data protection layer.
For AI agents
These docs are also available in machine-readable form for AI agents and LLMs:
- llms.txt: a page index with summaries
- llms-full.txt: the full text of every page in one file
- Each page’s raw Markdown: append
.mdto its URL (for example getting-started.md)
These files are generated automatically on every build and stay in sync with the content.